Intermediate layer optimization of HMAX model for face recognition

摘要

In this paper, we describe a quantitative model that accounts for the circuits and computations of the feed-forward path of the ventral stream of visual cortex. This model is consistent with a general theory of visual processing that extends the hierarchical model from primary to extra-striate visual areas. We implemented the Modified HMAX method, which has learning ability from C1 to S2 layer, and in order to S2 layer features optimization, we applied two clustering methods such as K-Means and Sequential Backward feature selection. After feature extraction, we used the K-nearest neighbor (KNN) and support vector machine (SVM) as classifiers. Experimental results have shown that applying the Sequential Backward feature selection in learning stage obtain higher recognition rate. The ORL database is exploited to test our approach. The experimental results showed the effectiveness of the system in terms of the recognition rate.